EVALUATION OF THE EFFECT OF MODELING UNCERTAINTIES ON THE SEISMIC RESPONSE OF EXISTING MASONRY BUILDINGS

An accurate seismic assessment of existing masonry buildings requires consideration of the sources of uncertainty and their effect on the seismic response. Among them, uncertainties related with modeling assumptions are definitely non-negligible. They include the choice of the modeling approach (continuum finite elements, equivalent frame, etc...), the analysis type (static/dynamic, linear/nonlinear) and the different options followed in defining the model. This work provides a quantitative evaluation, in probabilistic terms, of the effect of modeling uncertainties on the seismic response, in terms of the peak ground acceleration corresponding to the attainment of predefined limit states. Nonlinear static analysis was used, considered as the reference method to be adopted for the seismic assessment of existing buildings. The equivalent-frame macro-element approach was selected, being a satisfactory compromise between computational effort and accuracy in the results. Different possible choices regarding definition of the geometry of the equivalent frame, distribution of loads among the piers, distribution of loads on the floor systems, modeling of masonry spandrels, degree of coupling between orthogonal walls, definition of the cracked stiffness of structural elements, were considered. To quantify their effect on the response, a logic tree approach was followed, assigning a value of probability to each choice and hence obtaining the distribution of the acceleration corresponding to the selected limit states. This procedure was applied to a prototype building, for which a quantitative measure, in probabilistic terms, of the dispersion in the results due to the considered modeling uncertainties was evaluated.